JPH0228862A - Data transfer system in parallel logic type language - Google Patents

Abstract

PURPOSE:To decrease the frequency of data transfer through a network by sending data supply-drivingly from a producer process to a consumer process when the producer process writes data into a stream to a certain degree. CONSTITUTION:A producer process 4 is provided with a time monitoring means 41, a data termination detecting means 42 and a stream final detecting means 43. When the producer process 4 writes the data into the finite-length stream, the data are transferred to the side of the consumer process 5 when the time monitoring means 41 detects that the difference between time when the data are written and the value of a time area exceeds a fixed value set beforehand, the data terminal detecting means 42 detects the termination of the writing of the data and when the stream final detecting means 43 detects that it runs into the final of the stream.

Description

[Detailed Description of the Invention] [Summary] Regarding a data transfer method in a parallel logic language, for the purpose of improving the efficiency of communication between processors, a processor executing a producer process on the data sending side performs a data transfer method on the data receiving side. Assuming that the size of the stream, which is the bathophore memory used for communication with the processor that executes the consumer process, is finite length, there is an area in which the time at which the finite length stream is generated is written, and a symbol indicating the final element is written in the final element. In addition, the producer process includes a time monitoring means for monitoring that the difference between the time when the current stream is generated and the time at which new data is written has reached a predetermined time, and a time monitoring means for monitoring whether the data has reached the end. A data end detection means for detecting the end of data, a stream end detection means for detecting that the element position of the stream to which data is to be written is the last and extending it to a new finite length stream, and a time excess detection and end detection means by the time monitoring means. By detecting the end of the data by the controller and detecting the end of the stream by the stream end detector, the data is transferred to the consumer process in a supply-driven manner.

[Industrial application field]

The present invention relates to a data transfer method in a parallel logic language.

A parallel logic language is a language that is easy to describe in order to break down a problem into elements that can be processed in parallel so that a problem can be processed in parallel by multiple processors. The present invention relates to improving the efficiency of inter-processor communication when a program written in such a parallel logic language is processed by multiple processors connected through a network.

[Conventional technology]

In parallel logic languages, processes are the basic unit of processing.

Since many processes are generated during execution, it is necessary to manage them in a unified manner.

Therefore, as shown in FIG. 5, all processes are connected and managed in a queue called a ready queue.

A process is taken from this ready queue and executed only as long as it is executable. If you need to interrupt execution,
Connect to the end of the ready queue to pick up and execute another process.

Now, as shown in FIG. 6, process P is connected to process C, 1.2.3. Consider the case of sending data such as... Generally, a process that sends data like P is called a producer process, and a process that receives data like C is called a consumer process.

Such data exchange between processes (interprocess communication) is realized as shown in FIG.

That is, the producer process P writes data to a buffer called a stream, and the consumer process C reads it. Since processes P and C can run in parallel, C may try to read data before P can write it. So each element of the stream is initially initialized to "undefined", so
If C's read is faster than P's write, C will read "undefined." At that point, C remembers that "next time, just read this undefined part," interrupts the process, and connects it to the end of the ready queue. In this way, when a process interrupts its execution, it remembers where in the next stream to read or write from.

[Problem to be solved by the invention]

As shown in Figure 8, a parallel logic language characterized by inter-process communication as described above has loose coupling between processors,
Let us consider the case where the program is executed on a multiprocessor whose size can be changed arbitrarily (called scalable). 8th
In the figure, P and E are processor elements,
It consists of a central processing unit (CPU) and memory (MEM). Each PR is coupled via a network and cannot directly access the memory (MEM) of other PEs.

Now, as shown in FIG. 9, consider a case where a producer process P and a consumer process C are executed in different PRs. In order for consumer process C to receive data from producer process P, it is necessary to take the following steps.

■(PH2 side) Send a data transfer request message to the PH1 side.

■ (PE1 side) When a message is received, the data in the stream (data up to the undefined data) is sent to the PE2 side.

(2) (PE2 side) When receiving the data, PE2 copies the data to its own stream and starts executing process C.

(PE2 side) When process C is reading data and encounters an undefined condition, it sends a message requesting a subsequent data transfer.

Thereafter, this process is repeated until process C reaches the end of the data, that is, until process P finishes executing.

This method is fine for relatively small programs.

However, when larger-scale, general problems are programmed using this parallel logic language, the number of processes generated during execution increases even more, and each process cannot be executed continuously. That is, the process progresses by repeatedly writing or reading a small amount of data and then interrupting the process. When the producer process and the consumer process are in separate PEs, even if you send a message requesting data transfer via the network, there are often only 2 to 3 pieces of data in the stream, and the consumer Data transfer requests and data transfers must be made many times before the process ends. Compared to processing within a PE, data transfer via a network is mutually constrained and less efficient, and it is undesirable for this to occur frequently.

The problem to be solved by the present invention is to provide a data transfer method that eliminates such conventional problems.

[Means to solve the problem]

FIG. 1 is a principle block diagram of a data transfer method in a parallel logic language of the present invention.

In the figure, 1 is a processor element (PE), which consists of a CPU II and a memory (MEM) 12.

2 is also PR, CPU 21 and memory (MEM) 2
Consists of 2.

3 is a network, which connects PEs.

4 is a PF, and 1 is a producer process that is being executed.

5 is a consumer process being executed by PE2.

6 is a stream used by the producer process 4 to transfer data to the consumer process 5. The stream 6 has a finite length, and has an area in which the time at which the finite length stream was generated is written, and a symbol indicating the final element in the final element.

7 is a stream on the consumer process 5 side.

A time monitoring means 41 monitors when the difference between the time when the current stream is generated and the time when new data is written has reached a predetermined time.

42 is a data end detection means, which detects that the end of data has been reached.

43 is a stream end detection means which detects that the element position of the stream to which data is to be written is the last and extends it to a new finite length stream.

[For production]

According to the configuration of the present invention, when the producer process and the consumer process are executed on separate processors, the consumer process does not send a data transfer request message request-driven as is normally done. Rather, once the producer process has written a certain amount of data into the stream, the data is sent from the producer process to the consumer process in a supply-driven manner.

Therefore, in the present invention, as shown in FIG. 2, the stream of the producer process 4 is made to have a finite length. Then, for the last element of the stream, for example, "stream final" indicating the last element of the finite length stream is entered instead of "undefined". Additionally, a "time area" is provided in which the time when this finite length stream is generated is written.

In addition, the producer process 4 is provided with a time monitoring means 41, a data end detection means 42, and a stream final detection means 43, and when the producer process 4 writes data to a finite length stream, the time monitoring means 41 When the difference between the writing time and the value in the "time area" exceeds a predetermined constant value and a car is detected, ■ When the data end detection means 42 detects the end of data writing, and ■ Stream final detection. When the means 43 detects that the "end of stream" has been reached, it transfers the data to the consumer process 5 side.

By transferring data in a supply-driven manner in this way, the frequency of data transfer via the network can be reduced. The reason for providing the "time area" is to deal with the case where data does not accumulate in the stream for a long time.

〔Example〕

The present invention will be explained in more detail below with reference to embodiments shown in FIGS. 3 and 4.

FIG. 3 is a diagram illustrating the extension of a finite stream of a producer process in one embodiment of the present invention.

In the finite-length stream of the producer process P in this embodiment, as shown in FIG. Instead of "undefined", enter "undefined 2" to indicate the final value in the element.

When the producer process P encounters "Undefined 2", it allocates a new finite length stream and writes data there, as shown in Figure 3 fbl, and writes the data to "Undefined 2". Write a pointer to the data.

FIG. 4 is a flowchart showing the processing of the producer process in one embodiment of the present invention.

The operation of the producer process will be described below according to the steps of the flowchart.

■When the producer process reaches the end of the data, it writes a specific terminal symbol into the finite length stream and jumps to step ■. (The consumer process also reads the termination symbol to know the end of data transfer.) If it is not the end of the data, proceed to step (■).

■Determine whether the difference between the time t at which data is to be written and the time to (value of r time area) when the finite longevity dream is generated is greater than a certain time T, and if it is, jump to step ■; Otherwise, proceed to step ■.

It is determined whether the element of the stream to which the 0th order data is to be written is "undefined 2", and if it is "undefined 2", the process jumps to step (2); otherwise, the process goes to step (2).

■Write the data to the stream and return to step ■.

■Writes a terminal symbol, transfers the data to the consumer process, and finishes processing.

■The element of the stream to which data should be written is "Undefined 2"
”, and if not, proceed to step ■; if “undefined 2”, proceed to step ■.

■Write the data to the stream, transfer the data to the consumer process, update the time in the "time domain" to the current time, and return to step ■.

■ To extend the stream, as shown in Figure 3 (see BL), write the time in the "time area", write the data, transfer the data to the consumer process, and return to step (2).

〔Effect of the invention〕

As explained above, according to the present invention, the frequency of data transfer via a network can be reduced, and the effect of contributing to efficiency in executing parallel logic languages on loosely coupled multiprocessors is extremely large. It is.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of the principle of the present invention. FIG. 2 is a diagram showing communication between PEs according to the present invention. FIG. 3 is a diagram showing extension of a finite length stream in an embodiment of the present invention. Flowchart showing the processing of the producer process in one embodiment of the invention, FIG. 5 is a diagram showing a ready queue, FIG. 6 is a diagram showing the concept of inter-process communication, and FIG. 7 is a diagram showing realization of inter-process communication. , FIG. 8 is a diagram showing an example of a multiprocessor system, and FIG. 9 is a diagram showing inter-PE communication. In the drawing, 1.2 is a processor element, 3 is a network, 4 is a producer process, 5 is a consumer process, 6 and 7 are streams, 11 is a CPU,
12 is a memory (MEM), 41 is a time monitoring means, 42 is a data end detection means, and 43 is a stream end detection means. Patent Applicant Director of the Agency of Industrial Science and Technology Kozo Iizuka PE2 CY'I4 Consumer Alose Mata Stream Hon Noriaki 1 PL Interval Loss of Confidence Thousand τ Conversion Z Figure + Leap PE2 Consumer Process PE PP A f Meeting t X V';
EJ Man 9 Mouth

Claims (1)

[Claims] In a system in which a program written in a parallel logic language is executed on a multiprocessor in which a plurality of processors are connected via a network, a producer process (4) on the data sending side is executed. A stream (6), which is a buffer memory used for communication with a processor (2) that executes a consumer process (5) that receives data, has a finite length, and the finite length stream In addition to providing an area to write the generation time in (6) and a symbol indicating the final element in the final element, the producer process (4) is provided with an area to write the generation time and a time to write new data. a time monitoring means (41) for monitoring whether the difference has reached a predetermined time; a data end detection means (42) for detecting whether the data has reached the end; It is equipped with stream final detection means (43) for detecting that the stream is of a new finite length and extending it to a new finite length stream;
1), data end detection by the end detection means (42), and stream final detection means (43).
) in a supply-driven manner by detecting the end of the stream by
A data transfer method in a parallel logic language characterized by being configured to transfer data to a consumer process (5).